Abstract Details
Abstracts
Author: Pablo Rodriguez-Fernandez
Requested Type: Poster
Submitted: 2025-03-14 13:47:45
Co-authors: N. T. Howard, A. Martin-Sanabria, A. Ho, J. Pimentel-Aldaz, J. Hall, A. Saltzman and C. Holland
Contact Info:
MIT Plasma Science and Fusion Center
77 Mass Ave
Cambridge, 02139
United States
Abstract Text:
Core transport modeling of tokamak plasmas has become a critical tool to scope fusion power plants and plan experiments in upcoming burning plasmas. However, increasing fidelity of the turbulent transport models often comes with an associated increase in computational cost and additional challenges to reach convergence or steady-state conditions. Critical gradient, stiff transport and the uncertainty derived from either a limited time average in initial value solvers or discontinuities in eigenvalue solvers are usually the culprits for the high computational expense of transport simulations. This work aims to tackle these issues, with the goal of making physics-based transport simulations accessible to predict performance and enable their routine use for reactor design, as opposed to empirical scaling law extrapolations. To this end, refinements to the PORTALS framework [1] —performed under the Surrogate Models for Accurate and Rapid Transport Solutions (SMARTS) SciDAC partnership— are presented. Surrogate model selection, quantification of evaluation uncertainty, physics-guided optimization methods and custom convergence metrics are implemented in PORTALS, demonstrating the workflow can produce steady-state solutions using the TGLF quasilinear transport model [2] efficiently with minimal user input. PORTALS-TGLF has been used to produce databases of thousands of converged, steady-state transport simulations for numerical studies in experimentally-relevant plasma regimes, enabling studies that were only possible with neural network versions of turbulent transport models or with manual tuning and convergence checks.
This work was funded by US DoE DE-SC0024399.
[1] P. Rodriguez-Fernandez et al Nucl.Fusion 64 076034 (2024)
[2] G. M. Staebler et al Phys. Plasmas 14 055909 (2007)
Characterization: 6.0
Comments: